Copyright by International OCSCO World Press. All rights reserved.

2010 Research paper 9

VOLUME 41
ISSUES 1-2
July-August
2010
of Achievements in Materials
and Manufacturing Engineering
of Achievements in Materials
and Manufacturing Engineering
Frequency dependence of the
self-heating effect in polymer-based
composites
A. Katunin
a,
*, W. Hufenbach
b
, P. Kostka
b
, K. Holeczek
c

a
Department of Fundamentals of Machinery Design,
Silesian University of Technology, Konarskiego 18a, 44-100 Gliwice, Poland
b
Institute of Lightweight Engineering and Polymer Technology, Technische
Universitt Dresden, Holbeinstrae 3, 01307 Dresden, Germany
c
European Centre for Emerging Materials and Processes Dresden, Technische
Universitt Dresden, Marschnerstrae 39, 01307 Dresden, Germany
* Corresponding author: E-mail address: andrzej.katunin@polsl.pl
Received 30.05.2010; published in revised form 01.07.2010
Materials
ABSTRACT
Purpose: The self-heating effect caused by viscous energy dissipation in polymer-based composite structures
subjected to harmonic loads is considered to have a great influence on the residual life of the component. The
purpose of the conducted investigations is the determination of the dynamic mechanical behaviour of a polymer-
based composite material under different excitation frequencies and temperatures.
Design/methodology/approach: The dynamic mechanical analysis was employed for measurements of temperature
and frequency dependence of the complex rigidity parameters. Obtained loss rigidity curves for different load
frequencies enable the determination of the glass-transition temperatures and finally frequency-dependence of the loss
rigidity determined on the basis of the kinetic molecular theory and Williams-Landel-Ferry (WLF) hypothesis.
Findings: The dependency between glass-transition temperature and excitation frequency has been investigated.
The activation energy of the phase transition as well as the temperature dependence of the shift factor was calculated.
The glass-transition temperature and constants of WLF equation enable the determination of temperature and
frequency dependence of the loss rigidity according to the time-temperature superposition principle.
Research limitations/implications: The ranges of temperatures were limited to 30-150 C and excitation
frequencies to 1-200 Hz, the behaviour of the composite material outside these ranges can be estimated based on
the theoretical assumptions only. Obtained dependencies are correct only for linearly viscoelastic materials.
Practical implications: Obtained dependencies can be useful for estimation of the mechanical and thermal
degradation of polymer-based composites and can be subsequently applied for the determination of fatigue,
crack growth and residual life of composite structures.
Originality/value: The determination of temperature and frequency dependence of the loss rigidity gives an opportunity
to obtain the self-heating temperature distribution of the polymer-based composite structures under harmonic loading.
Keywords: Composites; Self-heating effect; Dynamic mechanical analysis
Reference to this paper should be given in the following way:
A. Katunin, W. Hufenbach, P. Kostka, K. Holeczek, Frequency dependence of the self-heating effect in polymer-
based composites, Journal of Achievements in Materials and Manufacturing Engineering 41/1-2 (2010) 9-15.
Research paper 10
Journal of Achievements in Materials and Manufacturing Engineering
A. Katunin, W. Hufenbach, P. Kostka, K. Holeczek
Volume 41 Issues 1-2 July-August 2010

1. Introduction

Constantly growing demands concerning safety, efficiency
and reliability of extremely-loaded components e.g. in automotive
and aerospace applications are resulting in increased interest in
the use of polymer-based composites. Particularly endless fibre-
reinforced composite materials with their outstanding, load-
adapted strength and stiffness are considered to be attractive for
the design of novel lightweight components with tailored
mechanical property profiles [1].
In automobile and aircraft structures, dynamic loads are
commonly present causing the fibre-reinforced composite
materials to exhibit their viscoelastic behaviour. This behaviour
causes energy dissipation, which is transformed into heat and
results in loss of process efficiency. Due to relatively small values
of heat transfer coefficients of amorphous polymers used in the
most matrix systems the dissipated heat accumulates in the
structure. When the dissipative heat and the thermal convection
conditions are balanced, a self-heating effect in the steady-state
can be observed, the temperature field is stabilised and
demonstrates a quasi-linear character against the time. In the case
of thermal instability the non steady-state of self-heating is
noticeable. Resulting from this effect temperature generation is
rapidly increasing above the glass-transition temperature, which
causes a material stiffness reduction. Such transitions may initiate
or accelerate damage processes and result in a malfunction of
polymer-based composite components [2]. Therefore, it is
necessary to determine the behaviour of these structures at various
operational temperatures and excitation frequencies in order to
describe the changing material behaviour. The constitutive model
of such loaded structures need to be changed from viscoelastic to
viscoelastoplastic or viscoplastic.
The viscoelastic behaviour of polymer-based composite
materials is determined by the matrix properties, due to low glass-
transition (-transition) temperature of most amorphous matrices
in comparison to the reinforcement material. Therefore the
properties of the reinforcing component can be neglected in
thermal problems.
Polymers used as composite matrices (e.g. epoxy resins)
exhibit viscoelastic response during oscillatory excitations and
their material parameters are both temperature- and frequency-
dependent. For the investigation of the loss rigidity of the
composite material as well as of its temperature and frequency
dependence, the dynamic mechanical analysis (DMA) was
selected. It provides the possibility to measure complex material
parameters in significant modes of oscillatory loading [3,4].
Applying a constant load or displacement, DMA gives also a
possibility to obtain creep (or relaxation) characteristics of
viscoelastic materials.
Theoretical investigations in the field of thermoviscoelasticity
were introduced by Karnaukhov, Senchenkov and Kozlov [5]. In
their works they propose governing equations of
thermoviscoelasticity in the form of complete and approximate
formulations. The self-heating effect was also investigated by
Dinzart and Molinari [6]. They described the self-heating effect in
axially loaded polymer-matrix beam subjected to pure bending
loads.
Katunins works were concentrated on the description of the
self-heating effect in polymer-based composite materials as well
as in the development of the fatigue models of polymer-based
composite plates. In [7] he presents an analytical model of the
cross-section of a rectangular plate based on complex rigidity.
The two-dimensional problem of steady-state self-heating effect
of the rectangular plate subjected to pure bending was discussed
in [8]. The self-heating effect in both steady- and non steady-
states and its influence on the crack initiation was investigated
using finite element analysis (FEA) [9]. Results on the crack
initiation problem in polymer-based composites based on FEA
were presented in many works (e.g. [10]), however they do not
consider an influence of the self-heating. The fatigue behaviour
considering self-heating effect was investigated using FEA; this
analysis leads to a formulation of analytical model concept,
describing the fatigue behaviour.
The aim of the investigations described in the current paper
was to determine the dependencies between the loss rigidity and
the temperature and excitation frequency of a rectangular plate
made of glass-fibre reinforced epoxy. Obtained dependencies
from the experimental results allow constructing the so-called
master curve, which presents the dependency of given complex
parameter simultaneously on the temperature and frequency. The
experimental investigations were conducted using the DMA-
thermoanalytical technique. Considering the above-mentioned
dependencies in the analytical model [8] it is possible to
determine temperature distribution caused by the self-heating
analytically.


2. Problem statement and theoretical
background


2.1. Motivation

The self-heating temperature distribution both in steady- and
non steady-state can be obtained theoretically using available
analytical models [5]. Using an approximate formulation of
thermoviscoelasticity, analytical models that include complex
parameters e. g. complex rigidity or complex modulus and their
components could be derived. In the problem of the structural
self-heating only the loss rigidity component is relevant, which
can be explained by the viscous nature of this parameter and its
influence on the hysteretic heating of the structure. According to
the governing equations of thermoviscoelasticity, the loss rigidity
is temperature- and frequency- dependent. Moreover, the value of
loss rigidity parameter can be determined only experimentally and
the DMA is considered to be the most adequate analysis for its
determination. Obtaining the above-mentioned dependencies
enable the formulation of the theoretical model that describes the
behaviour of the investigated materials at various operational
temperatures and excitation frequencies.


2.2. Dynamic mechanical analysis

Dynamic mechanical analysis is the combination of thermal
and rheological analysis, which provides corresponding values of
thermo-mechanical properties under dynamic loads, especially for
polymers and polymer-based composites. A dynamic mechanical
analyzer gives the possibility to determine the time-temperature-
or frequency-temperature-dependent complex parameters based
on phase shift between stress and strain (represented by tan ).
Due to various modes of operations like creep/relaxation,
isostrain, multi-frequency, multi-stress/strain, controlled
force/strain, the determination of -, - and -transitions as well
as thermal expansion coefficients is possible. The DMA can be
conducted in a broad temperature range (typically -150-600
o
C),
which enable the evaluation of material thermomechanical
properties for wide area of applications [11].


2.3. The constitutive model of
thermoviscoelasticity

Linearly viscoelastic behaviour of the polymer-based composite
materials can be expressed by Boltzmann-Volterra equation:
( ) ( ) ( )
|
|
.
|

\
|
H + =
}
t t o c d
1
0
t
t t
E
t
, (1)
where o(t) and c(t) are the stresses and deformations in an uniaxial
stressed state (e. g. creep) at the moment of measurement t, : is
time elapsed until the moment t, E is instantaneous Young's
modulus, H(t) is the relaxation kernel.
The stress tensor o
ij
is represented by the following formula:
ij ij ij ij
s o o o
3
1
=
, (2)
where s
ij
is the stress deviator and o
ij
is the Kroneckers delta. By
taking into consideration the temperature dependence and
modifying Eq. (1) due to the transversal isotropy of the composite
design, the deviatoric stress can be expressed as (cf. [8]):
( ) ( )
}
+ =
t
ij ij e ij
T t D T D s
0
d , 2 2 t c t c
. (3)
Here D
e
(T) is the temperature-dependent instantaneous rigidity of
the polymer-based composite material (where T denotes
temperature),
ij
c and
ij
c are the strain tensor component and its
first derivative, and D(t-:,T) is the time- and temperature-
dependent resolvent kernel.
According to the approximate formulation of the theory of
thermoviscoelasticity, the deviatoric stress can be expressed using
complex parameters. Equation (3) can therefore be presented as:
)] , (

) ( )[ (
max a a e ij
T D T D X w s e c + = , (4)
where w(X) is the deflection function dependent on the
coordinates vector X, T
a
is the cycle-averaged temperature, c is
the angular frequency, c
max
is the maximal deflection and
( )
a
T D ,

e
is the complex rigidity, which can be decomposed as:
( ) ( ) ( )
a a a
T D i T D T D , , ,

e e e ' ' + ' = , (5)

where D', D" are storage and loss rigidities respectively, which
physically denotes elastic and viscous response with respect to the
generalized Maxwell model of viscoelasticity:
( ) ( )
}

= '
0
d sin , , t t T t D T D
a a
e e e
, (6)
( ) ( )
}

= ' '
0
d cos , , t t T t D T D
a a
e e e
. (7)
The self-heating temperature can be determined from the heat
transfer equation:

( )
( ) | | ( ) t X Q t X
t
t X T
c
sh
, ,
,
= V V
c
c
u
. (8)
where c is the specific heat, o is the material density, is the
thermal conductivity and Q
sh
(X,t) is the source function, V is a
vector differential operator with respect to the coordinate vector
X. In this problem the source function is presented by the
dissipation energy based on the hysteresis loop area:
( )
}
=
t e
c o
e
t
2
0
d
2
, t t X Q
ij ij sh

. (9)
By solving (8) with applying thermal boundary conditions it is
possible to obtain the self-heating temperature distribution [7,8].
The investigated temperature distribution function according to
the Fouriers theorem can be presented in the form of infinite
double trigonometric series. For two-dimensional problems it can
be presented as:

( ) ( ) ( )
0
1 1
2
2
2
2
2
max
2 1
2
2 sin
1
2
2 sin
1
cos cos sin sin
, , 6 ,
u

u e c e u
+
|
|
.
|

\
|
+
|
|
.
|

\
|
+
|
|
.
|

\
|
+
' ' =

m n
n m
n
n
m
m
n m
n m
n m
a a
b l
b
v
l
u
D v u w v u
, (10)

where
m
and y
n
are subsequent roots of the boundary value
problem

l o = tan , b o = tan , (11)

u and v are Cartesian coordinates, l is the length of the plate, b is
the width of the plate and 0
0
is the ambient temperature.
As it can be noticed in (10), the only parameter that needs to be
determined is frequency- and temperature-dependent loss rigidity.
The determination of this parameter is possible in the series of
experiments using dynamic mechanical analyzer.


3. Experimental procedure

The specimens were manufactured from glass-fibre reinforced
polymer supplied by the company Epo GmbH in the form of
unidirectional preimpregnated fibres. The lay-up of the composite
1. Introduction
2.1. Motivation
2. Problem statement and
theoretical background
2.2. Dynamic mechanical analysis
11
Materials
Frequency dependence of the self-heating effect in polymer-based composites

1. Introduction

Constantly growing demands concerning safety, efficiency
and reliability of extremely-loaded components e.g. in automotive
and aerospace applications are resulting in increased interest in
the use of polymer-based composites. Particularly endless fibre-
reinforced composite materials with their outstanding, load-
adapted strength and stiffness are considered to be attractive for
the design of novel lightweight components with tailored
mechanical property profiles [1].
In automobile and aircraft structures, dynamic loads are
commonly present causing the fibre-reinforced composite
materials to exhibit their viscoelastic behaviour. This behaviour
causes energy dissipation, which is transformed into heat and
results in loss of process efficiency. Due to relatively small values
of heat transfer coefficients of amorphous polymers used in the
most matrix systems the dissipated heat accumulates in the
structure. When the dissipative heat and the thermal convection
conditions are balanced, a self-heating effect in the steady-state
can be observed, the temperature field is stabilised and
demonstrates a quasi-linear character against the time. In the case
of thermal instability the non steady-state of self-heating is
noticeable. Resulting from this effect temperature generation is
rapidly increasing above the glass-transition temperature, which
causes a material stiffness reduction. Such transitions may initiate
or accelerate damage processes and result in a malfunction of
polymer-based composite components [2]. Therefore, it is
necessary to determine the behaviour of these structures at various
operational temperatures and excitation frequencies in order to
describe the changing material behaviour. The constitutive model
of such loaded structures need to be changed from viscoelastic to
viscoelastoplastic or viscoplastic.
The viscoelastic behaviour of polymer-based composite
materials is determined by the matrix properties, due to low glass-
transition (-transition) temperature of most amorphous matrices
in comparison to the reinforcement material. Therefore the
properties of the reinforcing component can be neglected in
thermal problems.
Polymers used as composite matrices (e.g. epoxy resins)
exhibit viscoelastic response during oscillatory excitations and
their material parameters are both temperature- and frequency-
dependent. For the investigation of the loss rigidity of the
composite material as well as of its temperature and frequency
dependence, the dynamic mechanical analysis (DMA) was
selected. It provides the possibility to measure complex material
parameters in significant modes of oscillatory loading [3,4].
Applying a constant load or displacement, DMA gives also a
possibility to obtain creep (or relaxation) characteristics of
viscoelastic materials.
Theoretical investigations in the field of thermoviscoelasticity
were introduced by Karnaukhov, Senchenkov and Kozlov [5]. In
their works they propose governing equations of
thermoviscoelasticity in the form of complete and approximate
formulations. The self-heating effect was also investigated by
Dinzart and Molinari [6]. They described the self-heating effect in
axially loaded polymer-matrix beam subjected to pure bending
loads.
Katunins works were concentrated on the description of the
self-heating effect in polymer-based composite materials as well
as in the development of the fatigue models of polymer-based
composite plates. In [7] he presents an analytical model of the
cross-section of a rectangular plate based on complex rigidity.
The two-dimensional problem of steady-state self-heating effect
of the rectangular plate subjected to pure bending was discussed
in [8]. The self-heating effect in both steady- and non steady-
states and its influence on the crack initiation was investigated
using finite element analysis (FEA) [9]. Results on the crack
initiation problem in polymer-based composites based on FEA
were presented in many works (e.g. [10]), however they do not
consider an influence of the self-heating. The fatigue behaviour
considering self-heating effect was investigated using FEA; this
analysis leads to a formulation of analytical model concept,
describing the fatigue behaviour.
The aim of the investigations described in the current paper
was to determine the dependencies between the loss rigidity and
the temperature and excitation frequency of a rectangular plate
made of glass-fibre reinforced epoxy. Obtained dependencies
from the experimental results allow constructing the so-called
master curve, which presents the dependency of given complex
parameter simultaneously on the temperature and frequency. The
experimental investigations were conducted using the DMA-
thermoanalytical technique. Considering the above-mentioned
dependencies in the analytical model [8] it is possible to
determine temperature distribution caused by the self-heating
analytically.


2. Problem statement and theoretical
background


2.1. Motivation

The self-heating temperature distribution both in steady- and
non steady-state can be obtained theoretically using available
analytical models [5]. Using an approximate formulation of
thermoviscoelasticity, analytical models that include complex
parameters e. g. complex rigidity or complex modulus and their
components could be derived. In the problem of the structural
self-heating only the loss rigidity component is relevant, which
can be explained by the viscous nature of this parameter and its
influence on the hysteretic heating of the structure. According to
the governing equations of thermoviscoelasticity, the loss rigidity
is temperature- and frequency- dependent. Moreover, the value of
loss rigidity parameter can be determined only experimentally and
the DMA is considered to be the most adequate analysis for its
determination. Obtaining the above-mentioned dependencies
enable the formulation of the theoretical model that describes the
behaviour of the investigated materials at various operational
temperatures and excitation frequencies.


2.2. Dynamic mechanical analysis

Dynamic mechanical analysis is the combination of thermal
and rheological analysis, which provides corresponding values of
thermo-mechanical properties under dynamic loads, especially for
polymers and polymer-based composites. A dynamic mechanical
analyzer gives the possibility to determine the time-temperature-
or frequency-temperature-dependent complex parameters based
on phase shift between stress and strain (represented by tan ).
Due to various modes of operations like creep/relaxation,
isostrain, multi-frequency, multi-stress/strain, controlled
force/strain, the determination of -, - and -transitions as well
as thermal expansion coefficients is possible. The DMA can be
conducted in a broad temperature range (typically -150-600
o
C),
which enable the evaluation of material thermomechanical
properties for wide area of applications [11].


2.3. The constitutive model of
thermoviscoelasticity

Linearly viscoelastic behaviour of the polymer-based composite
materials can be expressed by Boltzmann-Volterra equation:
( ) ( ) ( )
|
|
.
|

\
|
H + =
}
t t o c d
1
0
t
t t
E
t
, (1)
where o(t) and c(t) are the stresses and deformations in an uniaxial
stressed state (e. g. creep) at the moment of measurement t, : is
time elapsed until the moment t, E is instantaneous Young's
modulus, H(t) is the relaxation kernel.
The stress tensor o
ij
is represented by the following formula:
ij ij ij ij
s o o o
3
1
=
, (2)
where s
ij
is the stress deviator and o
ij
is the Kroneckers delta. By
taking into consideration the temperature dependence and
modifying Eq. (1) due to the transversal isotropy of the composite
design, the deviatoric stress can be expressed as (cf. [8]):
( ) ( )
}
+ =
t
ij ij e ij
T t D T D s
0
d , 2 2 t c t c
. (3)
Here D
e
(T) is the temperature-dependent instantaneous rigidity of
the polymer-based composite material (where T denotes
temperature),
ij
c and
ij
c are the strain tensor component and its
first derivative, and D(t-:,T) is the time- and temperature-
dependent resolvent kernel.
According to the approximate formulation of the theory of
thermoviscoelasticity, the deviatoric stress can be expressed using
complex parameters. Equation (3) can therefore be presented as:
)] , (

) ( )[ (
max a a e ij
T D T D X w s e c + = , (4)
where w(X) is the deflection function dependent on the
coordinates vector X, T
a
is the cycle-averaged temperature, c is
the angular frequency, c
max
is the maximal deflection and
( )
a
T D ,

e
is the complex rigidity, which can be decomposed as:
( ) ( ) ( )
a a a
T D i T D T D , , ,

e e e ' ' + ' = , (5)

where D', D" are storage and loss rigidities respectively, which
physically denotes elastic and viscous response with respect to the
generalized Maxwell model of viscoelasticity:
( ) ( )
}

= '
0
d sin , , t t T t D T D
a a
e e e
, (6)
( ) ( )
}

= ' '
0
d cos , , t t T t D T D
a a
e e e
. (7)
The self-heating temperature can be determined from the heat
transfer equation:

( )
( ) | | ( ) t X Q t X
t
t X T
c
sh
, ,
,
= V V
c
c
u
. (8)
where c is the specific heat, o is the material density, is the
thermal conductivity and Q
sh
(X,t) is the source function, V is a
vector differential operator with respect to the coordinate vector
X. In this problem the source function is presented by the
dissipation energy based on the hysteresis loop area:
( )
}
=
t e
c o
e
t
2
0
d
2
, t t X Q
ij ij sh

. (9)
By solving (8) with applying thermal boundary conditions it is
possible to obtain the self-heating temperature distribution [7,8].
The investigated temperature distribution function according to
the Fouriers theorem can be presented in the form of infinite
double trigonometric series. For two-dimensional problems it can
be presented as:

( ) ( ) ( )
0
1 1
2
2
2
2
2
max
2 1
2
2 sin
1
2
2 sin
1
cos cos sin sin
, , 6 ,
u

u e c e u
+
|
|
.
|

\
|
+
|
|
.
|

\
|
+
|
|
.
|

\
|
+
' ' =

m n
n m
n
n
m
m
n m
n m
n m
a a
b l
b
v
l
u
D v u w v u
, (10)

where
m
and y
n
are subsequent roots of the boundary value
problem

l o = tan , b o = tan , (11)

u and v are Cartesian coordinates, l is the length of the plate, b is
the width of the plate and 0
0
is the ambient temperature.
As it can be noticed in (10), the only parameter that needs to be
determined is frequency- and temperature-dependent loss rigidity.
The determination of this parameter is possible in the series of
experiments using dynamic mechanical analyzer.


3. Experimental procedure

The specimens were manufactured from glass-fibre reinforced
polymer supplied by the company Epo GmbH in the form of
unidirectional preimpregnated fibres. The lay-up of the composite
2.3. The constitutive model
of thermoviscoelasticity
3. Experimental procedure
Research paper 12
Journal of Achievements in Materials and Manufacturing Engineering
A. Katunin, W. Hufenbach, P. Kostka, K. Holeczek
Volume 41 Issues 1-2 July-August 2010

Table 1.
Material properties of the single UD-layer [18]
E
1
[GPa] E
2
[GPa] G
12
[GPa]
12
[-] [kg/m
3
]
38.283 10.141 3.533 0.366 1794

material, [0/ 60/ -60/ -60/ 60/ 0]
s
, was selected in order to achieve
transversal isotropic properties. Table 1 shows the material
properties of the single UD-layer of the composite material. The
dimensions of the specimen are: width W=17.50.1 mm, width
H=10.00.1 mm and thickness T=2.50.05 mm (Fig.1.).



Fig. 1. Geometrical configuration of the test specimen

Dynamic mechanical analysis was conducted on the DMA-
Q800 analyser of TA Instruments. The specimens were fastened
in a single-cantilever clamp (Fig. 2.). The temperature range of
the investigations in multi-frequency-strain operational mode was
set to 30-150 C and the corresponding heating rate to 3 K/min
(conform to the requirements described in [12]). Tests in the
above-mentioned configuration were performed for following
excitation frequencies: 1, 2, 5, 10, 20, 50, 75, 100, 125, 150, 175
and 200 Hz. For each case, three specimens were tested in order
to average the obtained results and hence to minimise the
statistical deviation. The investigations were performed under
constant excitation amplitude of 20 m. Following parameters
were measured and stored during the experimental investigations:
temperature, frequency, storage and loss moduli. Selected results
of conducted DMA measurements are presented in Fig. 3.




Fig. 2. Experimental setup of the DMA Analyser




Fig. 3. Temperatures scan at frequency 10 Hz

Single
cantilever
clamping
Test specimen
Acquired characteristic material parameters have been applied for
determination of the glass-transition temperatures, T
g
based on the
peak value of tan or loss modulus. Since the main focus of this
study lies on the loss modulus (which can be used together with
geometrical parameters for the calculation of the loss rigidity), this
parameter was selected as the basis for the determination of the
glass-transition temperatures. Results obtained from experimental
investigations for different frequencies have been stored, and
subsequently analysed in numerical computing environment.


4. Results and discussion

Averaged glass-transition temperatures for above-mentioned
frequency values were determined. They show that T
g
increases
with the increase of frequency. It is found that the relation
between T
g
and frequency follows the kinetic molecular theory
has an exponential character and can be expressed by the
Arrhenius equation (cf. [13]):

|
|
.
|

\
|
=
0
0
exp
RT
E
f f
a
, (12)

where f
0
is the empirically determined pre-exponential factor, E
a

is the activation energy, R is the universal gas constant (8.314472
J/Kmol) and T
0
is the reference temperature, in the present study
T
0
= T
g
. Fig. 4 shows the relationship between ln(f) and 1/T
g
.



Fig. 4. Arrhenius relationship

Obtained data was fitted using linear regression algorithm. The R
2

is equal 0.984. From the slope, an activation energy E
a
of 395.37
kJmol
-1
can be calculated. Using this value it is possible to
determine shift factors in function of temperature using modified
Arrhenius equation (13) [14]. The shift factors are necessary for
the construction of a master curve for the loss rigidity. According
to the variability of T
g
the averaged value of T
g
was applied for
the reference temperature T
0
= 105 C.

( )
|
|
.
|

\
|
=
0
1 1
303 . 2
ln
T T R
E
a
a
T
, (13)
where a
T
is the shift factor and T is the experimental temperature.
The relationship of ln(a
T
) and T-T
0
is presented in Fig. 5.
Results show, that the laminate reveals linear viscoelastic
behaviour, so that the time-temperature superposition (TTS)
principle can be applied in the investigated case. Following TTS
the behaviour of linearly viscoelastic polymeric materials at the
temperature T
0
could be related to that at another temperature T
by changing the frequency scale [14]:

( ) ( ) T D T a D
T
, ,
0
e e ' ' = ' ' , (14)
with = 2f.
The obtained values of shift factor allow to apply the
Williams-Landel-Ferry (WLF) empirical hypothesis [15] for
determining constants of WLF equation (15) proposed by Ferry in
[16] using curve fitting.

( )
( )
0 2
0 1
ln
T T C
T T C
a
T
+

=
, (15)
Based on the measured values of the loss rigidity and the values
of a
T
it is possible to construct the master curve of the loss rigidity
of the laminate. Therefore it is necessary to present the loss
rigidity sequences in frequency domain (Fig. 6.) and to multiply
frequency values by the shift factors. During this operation
sequences are translated by horizontal shifting. Depending on the
value of T
g
sequences are shifted to the left (when T
g
is less than
the reference temperature) or to the right (when T
g
is higher than
the reference temperature). The obtained master curve is
presented in Fig. 7.



Fig. 5. Shift factor vs. temperature difference for the reference
temperature of 105 C

Obtained values of C
1
and C
2
for the investigated laminate are
54.4745 and 377.288 (C or K) respectively. If condition of T
0
= T
g

is satisfied the values of these constants can be assumed as
universal ones (equal 17.44 and 51.6 respectively) for most
polymers [17], but for some materials they may differ, which is
confirmed, e.g. in [14]. In fact, there is a large deviation of these
constants from polymer to polymer and it has been observed in
many polymer systems. It is often attributed to specific character of
the polymer system [18]. In the investigated study WLF equation
constants does not vary with respect to T
g
for different frequencies.
The maximal differences of values of C
1
and C
2
in comparison with
values obtained for reference temperatures of specified cases (see
Fig. 4) are -0.16 % and -0.17 % respectively.
13
Materials
Frequency dependence of the self-heating effect in polymer-based composites

Table 1.
Material properties of the single UD-layer [18]
E
1
[GPa] E
2
[GPa] G
12
[GPa]
12
[-] [kg/m
3
]
38.283 10.141 3.533 0.366 1794

material, [0/ 60/ -60/ -60/ 60/ 0]
s
, was selected in order to achieve
transversal isotropic properties. Table 1 shows the material
properties of the single UD-layer of the composite material. The
dimensions of the specimen are: width W=17.50.1 mm, width
H=10.00.1 mm and thickness T=2.50.05 mm (Fig.1.).



Fig. 1. Geometrical configuration of the test specimen

Dynamic mechanical analysis was conducted on the DMA-
Q800 analyser of TA Instruments. The specimens were fastened
in a single-cantilever clamp (Fig. 2.). The temperature range of
the investigations in multi-frequency-strain operational mode was
set to 30-150 C and the corresponding heating rate to 3 K/min
(conform to the requirements described in [12]). Tests in the
above-mentioned configuration were performed for following
excitation frequencies: 1, 2, 5, 10, 20, 50, 75, 100, 125, 150, 175
and 200 Hz. For each case, three specimens were tested in order
to average the obtained results and hence to minimise the
statistical deviation. The investigations were performed under
constant excitation amplitude of 20 m. Following parameters
were measured and stored during the experimental investigations:
temperature, frequency, storage and loss moduli. Selected results
of conducted DMA measurements are presented in Fig. 3.




Fig. 2. Experimental setup of the DMA Analyser




Fig. 3. Temperatures scan at frequency 10 Hz

Single
cantilever
clamping
Test specimen
Acquired characteristic material parameters have been applied for
determination of the glass-transition temperatures, T
g
based on the
peak value of tan or loss modulus. Since the main focus of this
study lies on the loss modulus (which can be used together with
geometrical parameters for the calculation of the loss rigidity), this
parameter was selected as the basis for the determination of the
glass-transition temperatures. Results obtained from experimental
investigations for different frequencies have been stored, and
subsequently analysed in numerical computing environment.


4. Results and discussion

Averaged glass-transition temperatures for above-mentioned
frequency values were determined. They show that T
g
increases
with the increase of frequency. It is found that the relation
between T
g
and frequency follows the kinetic molecular theory
has an exponential character and can be expressed by the
Arrhenius equation (cf. [13]):

|
|
.
|

\
|
=
0
0
exp
RT
E
f f
a
, (12)

where f
0
is the empirically determined pre-exponential factor, E
a

is the activation energy, R is the universal gas constant (8.314472
J/Kmol) and T
0
is the reference temperature, in the present study
T
0
= T
g
. Fig. 4 shows the relationship between ln(f) and 1/T
g
.



Fig. 4. Arrhenius relationship

Obtained data was fitted using linear regression algorithm. The R
2

is equal 0.984. From the slope, an activation energy E
a
of 395.37
kJmol
-1
can be calculated. Using this value it is possible to
determine shift factors in function of temperature using modified
Arrhenius equation (13) [14]. The shift factors are necessary for
the construction of a master curve for the loss rigidity. According
to the variability of T
g
the averaged value of T
g
was applied for
the reference temperature T
0
= 105 C.

( )
|
|
.
|

\
|
=
0
1 1
303 . 2
ln
T T R
E
a
a
T
, (13)
where a
T
is the shift factor and T is the experimental temperature.
The relationship of ln(a
T
) and T-T
0
is presented in Fig. 5.
Results show, that the laminate reveals linear viscoelastic
behaviour, so that the time-temperature superposition (TTS)
principle can be applied in the investigated case. Following TTS
the behaviour of linearly viscoelastic polymeric materials at the
temperature T
0
could be related to that at another temperature T
by changing the frequency scale [14]:

( ) ( ) T D T a D
T
, ,
0
e e ' ' = ' ' , (14)
with = 2f.
The obtained values of shift factor allow to apply the
Williams-Landel-Ferry (WLF) empirical hypothesis [15] for
determining constants of WLF equation (15) proposed by Ferry in
[16] using curve fitting.

( )
( )
0 2
0 1
ln
T T C
T T C
a
T
+

=
, (15)
Based on the measured values of the loss rigidity and the values
of a
T
it is possible to construct the master curve of the loss rigidity
of the laminate. Therefore it is necessary to present the loss
rigidity sequences in frequency domain (Fig. 6.) and to multiply
frequency values by the shift factors. During this operation
sequences are translated by horizontal shifting. Depending on the
value of T
g
sequences are shifted to the left (when T
g
is less than
the reference temperature) or to the right (when T
g
is higher than
the reference temperature). The obtained master curve is
presented in Fig. 7.



Fig. 5. Shift factor vs. temperature difference for the reference
temperature of 105 C

Obtained values of C
1
and C
2
for the investigated laminate are
54.4745 and 377.288 (C or K) respectively. If condition of T
0
= T
g

is satisfied the values of these constants can be assumed as
universal ones (equal 17.44 and 51.6 respectively) for most
polymers [17], but for some materials they may differ, which is
confirmed, e.g. in [14]. In fact, there is a large deviation of these
constants from polymer to polymer and it has been observed in
many polymer systems. It is often attributed to specific character of
the polymer system [18]. In the investigated study WLF equation
constants does not vary with respect to T
g
for different frequencies.
The maximal differences of values of C
1
and C
2
in comparison with
values obtained for reference temperatures of specified cases (see
Fig. 4) are -0.16 % and -0.17 % respectively.
4. Results and discussion
Research paper 14
Journal of Achievements in Materials and Manufacturing Engineering
A. Katunin, W. Hufenbach, P. Kostka, K. Holeczek
Volume 41 Issues 1-2 July-August 2010



Fig. 6. Frequency dependency of the loss rigidity at different
isothermal temperatures



Fig. 7. Master curve of loss rigidity with respect to the reference
temperature of 105 C

The shape of the obtained master curve is in the good
agreement with literature results for such materials [14]. The
presented algorithm could be useful to determine viscoelastic
parameters in form of Pronys series (16), which is based on the
generalized Maxwell rheological model and often used in FEM
commercial software:

( )

=
|
.
|

\
|
+ =
N
n
n
n
ij
e
ij ij
t
D D t D
1
exp
t
. (16)

During the experiment it was noticed, that the glass-transition
temperature could be determined from the relation between force
and temperature. The exemplary relation is given by Fig. 8.
Note, that the relation obtained in Fig.8 is equivalent with the
relation of the storage moduli versus temperature (compare with
Fig. 3). The relation presented in Fig. 8 could be obtained without
the necessity of the dynamic mechanical analysis, which is often
expensive or not available. According to small differences
between above-mentioned techniques, the precise force
measurements in function of the temperature could be applied for
T
g
determination as well. The method of glass-transition
temperature evaluation based on the analysis of storage modulus
versus temperature relation is the most imprecise (usually T
g
is
determined based on the loss modulus or tan relations).
Therefore the proposed method will be imprecise too. However,
in most engineering applications such slight imprecision is
negligible.




Fig. 8. Relation between force and temperature for 1 Hz


5. Conclusions

In this work the influence of temperature and frequency on
the loss rigidity for linearly viscoelastic laminate was determined.
DMA experiments conducted good agreement of measurements
with theoretical models. Proposed algorithm of above-mentioned
relations determination could be applied for evaluation of
complex parameters of linear viscoelastic materials as well.
Constants of the WLF equation were determined and were vary to
the universal ones. Therefore, it is necessary to construct the
shift factor vs. temperature curve to obtain adequate values of
them and to investigate their dependence on the reference
temperature, which may change the values of constants, especially
for C
2
. Obtaining the temperature and frequency dependence of
the loss rigidity and taking into consideration the TTS principle it
is possible to fulfill the analytical models of the self-heating
temperature distribution and the character of increasing proposed
in [7,8]. The proposed technique of the glass-transition
temperature determination based on force measurements could be
low-costly in comparison with DMA.
The self-heating effect could be used in problems of the
damage detection and evaluation in polymer-based composite
structures. Several research on this field was provided by
scientific group of Prof. Wrbel [20-22]. In [20] the authors used
pulsed infrared thermography for detection of defects. This
method is based on the short-time heating of the composite
structure by the internal heat source and the analysis of the
obtained thermograms. The self-heating effect allows the damage
detection without the heat source. The heating of the structure
could be provided by choosing appropriate loading parameters.
In further research it is necessary to conduct the experiment of
the self-heating of laminate for verifying theoretical studies. The
temperature rises will change the frequency spectrum of cyclically
loaded composite plates due to the stress relaxation processes in
the material [23], thus it is necessary to determine the dependence
of the modal characteristics in a frequency domain on the
temperature, which gives a possibility to predict shifting of the
structures resonant frequencies.


Acknowledgements

The particular financial support of the Polish Ministry of Science
and Higher Education in the research grant No. N N504 282137 is
gratefully acknowledged.


References

[1] W. Hufenbach, F. Adam, M. Zichner, M. Krahl,
Development and production of composites in multimaterial
design, Proceedings of the 2
th
Aachen-Dresden International
Textile Conference, Dresden, 2008.
[2] A. Katunin, W. Moczulski, The conception of a
methodology of degradation degree evaluation of laminates,
Maintenance and Reliability 41 (2009) 33-38.
[3] K.P. Menard (Ed.), Dynamic mechanical analysis: A
practical introduction, CRC Press, Washington D.C., 1999.
[4] P. Postawa, A. Szarek, J. Kokszul, DMTA method in
determining strength parameters of acrylic cements,
Archives of Materials Science and Engineering 28 (2007)
309-312.
[5] I.K. Senchenkov, V.G. Karnaukhov, V.I. Kozlov, Toward a
theory of governing equations of thermoviscoelasticity for
periodic deformation, Trans.: Prikladnaya Mekhanika 22
(1986) 97-104.
[6] F. Dinzart, A. Molinari, R. Herbach, Thermomechanical
response of viscoelastic beam under cyclic axial loading;
self-heating and thermal failure, Archives of Mechanics 60
(2008) 59-85.
[7] A. Katunin, Self-heating effect in laminate plates during
harmonic forced loading, Scientific Problems of Machine
Operation and Maintenance 44 (2009) 73-84.
[8] A. Katunin, Analytical model of the self-heating effect in
polymeric laminated rectangular plates during bending
harmonic loading, Maintenance and Reliability 48 (2010)
91-101 (in Polish).
[9] A. Katunin, Influence of self-heating temperature on the
fatigue strength of a plate made of laminated polymeric
composite, International Applied Mechanics 45 (2009) 342.
[10] T. Da Silva Botelho, N. Isac, E. Bayraktar, A comparative
study on the damage initiation mechanism of elastomeric
composites, Archives of Computational Materials Science
and Surface Engineering 1 (2009) 112-119.
[11] TA Instruments, 2980 Dynamic Mechanical Analyzer
Operators Manual, 1997.
[12] EN-ISO 6721-1:2002. Plastics Determination of dynamic
mechanical properties Part 1: General principles.
[13] L.L. Gao, X. Chen, S.B. Zhang, H. Gao, Mechanical
properties of anisotropic conductive film with strain rate and
temperature, Material Science and Engineering A 513-514
(2009) 216-221.
[14] Y. He, Thermomechanical and viscoelastic behaviour of a
non-flow underfill material for flip-chip applications,
Thermochimica Acta 439 (2005) 127-134.
[15] M.L. Williams, R.F. Landel, J.D. Ferry, The temperature
dependence of relaxation mechanisms in amorphous
polymers and other glass-forming liquids, Journal of the
American Chemical Society 77 (1955) 3701-3707.
[16] J.D. Ferry, Viscoelastic properties of polymers, Third
Edition, Wiley, New York, 1980.
[17] P.C. Heimenz, T.P. Lorge, Polymer chemistry, Boca Raton,
2007.
[18] C.A. Angell, Why C
1
= 16-17 in the WLF equation is
physical - and the fragility of polymers, Polymer 26 (1997)
6261-6266.
[19] S. Dbrowa, Modelling of the degradation process of the
composite rotor material and its influence on the dynamical
rotor behaviour, MSc Thesis, Silesian University of
Technology, 2006 (in Polish).
[20] M. Szczepanik, J. Stabik, G. Wrbel, . Wierzbicki,
Detecting of defects in polymeric materials using pulsed
infrared thermography, Archives of Materials Science and
Engineering 30 (2008) 29-32.
[21] J. Kaczmarczyk, M. Rojek, G. Wrbel, J. Stabik, A model of
heat transfer in composites subjected to thermographic
testing, Archives of Materials Science and Engineering 31
(2008) 105-108.
[22] G. Wrbel, Z.M. Rdzawski, G. Muzia, S. Pawlak,
Quantitative analysis of the fiber content distribution in CFRP
composites using thermal non-destructive testing, Archives of
Materials Science and Engineering 41 (2010) 28-36.
[23] B. Diveev, A. Smolskyy, M. Sukhorolskyy, Dynamic
rigidity and loss factor prediction for composite layered
panel, Archives of Materials Science and Engineering 31
(2008) 45-48.

5. Conclusions
15 READING DIRECT: www.journalamme.org
Materials



Fig. 6. Frequency dependency of the loss rigidity at different
isothermal temperatures



Fig. 7. Master curve of loss rigidity with respect to the reference
temperature of 105 C

The shape of the obtained master curve is in the good
agreement with literature results for such materials [14]. The
presented algorithm could be useful to determine viscoelastic
parameters in form of Pronys series (16), which is based on the
generalized Maxwell rheological model and often used in FEM
commercial software:

( )

=
|
.
|

\
|
+ =
N
n
n
n
ij
e
ij ij
t
D D t D
1
exp
t
. (16)

During the experiment it was noticed, that the glass-transition
temperature could be determined from the relation between force
and temperature. The exemplary relation is given by Fig. 8.
Note, that the relation obtained in Fig.8 is equivalent with the
relation of the storage moduli versus temperature (compare with
Fig. 3). The relation presented in Fig. 8 could be obtained without
the necessity of the dynamic mechanical analysis, which is often
expensive or not available. According to small differences
between above-mentioned techniques, the precise force
measurements in function of the temperature could be applied for
T
g
determination as well. The method of glass-transition
temperature evaluation based on the analysis of storage modulus
versus temperature relation is the most imprecise (usually T
g
is
determined based on the loss modulus or tan relations).
Therefore the proposed method will be imprecise too. However,
in most engineering applications such slight imprecision is
negligible.




Fig. 8. Relation between force and temperature for 1 Hz


5. Conclusions

In this work the influence of temperature and frequency on
the loss rigidity for linearly viscoelastic laminate was determined.
DMA experiments conducted good agreement of measurements
with theoretical models. Proposed algorithm of above-mentioned
relations determination could be applied for evaluation of
complex parameters of linear viscoelastic materials as well.
Constants of the WLF equation were determined and were vary to
the universal ones. Therefore, it is necessary to construct the
shift factor vs. temperature curve to obtain adequate values of
them and to investigate their dependence on the reference
temperature, which may change the values of constants, especially
for C
2
. Obtaining the temperature and frequency dependence of
the loss rigidity and taking into consideration the TTS principle it
is possible to fulfill the analytical models of the self-heating
temperature distribution and the character of increasing proposed
in [7,8]. The proposed technique of the glass-transition
temperature determination based on force measurements could be
low-costly in comparison with DMA.
The self-heating effect could be used in problems of the
damage detection and evaluation in polymer-based composite
structures. Several research on this field was provided by
scientific group of Prof. Wrbel [20-22]. In [20] the authors used
pulsed infrared thermography for detection of defects. This
method is based on the short-time heating of the composite
structure by the internal heat source and the analysis of the
obtained thermograms. The self-heating effect allows the damage
detection without the heat source. The heating of the structure
could be provided by choosing appropriate loading parameters.
In further research it is necessary to conduct the experiment of
the self-heating of laminate for verifying theoretical studies. The
temperature rises will change the frequency spectrum of cyclically
loaded composite plates due to the stress relaxation processes in
the material [23], thus it is necessary to determine the dependence
of the modal characteristics in a frequency domain on the
temperature, which gives a possibility to predict shifting of the
structures resonant frequencies.


Acknowledgements

The particular financial support of the Polish Ministry of Science
and Higher Education in the research grant No. N N504 282137 is
gratefully acknowledged.


References

[1] W. Hufenbach, F. Adam, M. Zichner, M. Krahl,
Development and production of composites in multimaterial
design, Proceedings of the 2
th
Aachen-Dresden International
Textile Conference, Dresden, 2008.
[2] A. Katunin, W. Moczulski, The conception of a
methodology of degradation degree evaluation of laminates,
Maintenance and Reliability 41 (2009) 33-38.
[3] K.P. Menard (Ed.), Dynamic mechanical analysis: A
practical introduction, CRC Press, Washington D.C., 1999.
[4] P. Postawa, A. Szarek, J. Kokszul, DMTA method in
determining strength parameters of acrylic cements,
Archives of Materials Science and Engineering 28 (2007)
309-312.
[5] I.K. Senchenkov, V.G. Karnaukhov, V.I. Kozlov, Toward a
theory of governing equations of thermoviscoelasticity for
periodic deformation, Trans.: Prikladnaya Mekhanika 22
(1986) 97-104.
[6] F. Dinzart, A. Molinari, R. Herbach, Thermomechanical
response of viscoelastic beam under cyclic axial loading;
self-heating and thermal failure, Archives of Mechanics 60
(2008) 59-85.
[7] A. Katunin, Self-heating effect in laminate plates during
harmonic forced loading, Scientific Problems of Machine
Operation and Maintenance 44 (2009) 73-84.
[8] A. Katunin, Analytical model of the self-heating effect in
polymeric laminated rectangular plates during bending
harmonic loading, Maintenance and Reliability 48 (2010)
91-101 (in Polish).
[9] A. Katunin, Influence of self-heating temperature on the
fatigue strength of a plate made of laminated polymeric
composite, International Applied Mechanics 45 (2009) 342.
[10] T. Da Silva Botelho, N. Isac, E. Bayraktar, A comparative
study on the damage initiation mechanism of elastomeric
composites, Archives of Computational Materials Science
and Surface Engineering 1 (2009) 112-119.
[11] TA Instruments, 2980 Dynamic Mechanical Analyzer
Operators Manual, 1997.
[12] EN-ISO 6721-1:2002. Plastics Determination of dynamic
mechanical properties Part 1: General principles.
[13] L.L. Gao, X. Chen, S.B. Zhang, H. Gao, Mechanical
properties of anisotropic conductive film with strain rate and
temperature, Material Science and Engineering A 513-514
(2009) 216-221.
[14] Y. He, Thermomechanical and viscoelastic behaviour of a
non-flow underfill material for flip-chip applications,
Thermochimica Acta 439 (2005) 127-134.
[15] M.L. Williams, R.F. Landel, J.D. Ferry, The temperature
dependence of relaxation mechanisms in amorphous
polymers and other glass-forming liquids, Journal of the
American Chemical Society 77 (1955) 3701-3707.
[16] J.D. Ferry, Viscoelastic properties of polymers, Third
Edition, Wiley, New York, 1980.
[17] P.C. Heimenz, T.P. Lorge, Polymer chemistry, Boca Raton,
2007.
[18] C.A. Angell, Why C
1
= 16-17 in the WLF equation is
physical - and the fragility of polymers, Polymer 26 (1997)
6261-6266.
[19] S. Dbrowa, Modelling of the degradation process of the
composite rotor material and its influence on the dynamical
rotor behaviour, MSc Thesis, Silesian University of
Technology, 2006 (in Polish).
[20] M. Szczepanik, J. Stabik, G. Wrbel, . Wierzbicki,
Detecting of defects in polymeric materials using pulsed
infrared thermography, Archives of Materials Science and
Engineering 30 (2008) 29-32.
[21] J. Kaczmarczyk, M. Rojek, G. Wrbel, J. Stabik, A model of
heat transfer in composites subjected to thermographic
testing, Archives of Materials Science and Engineering 31
(2008) 105-108.
[22] G. Wrbel, Z.M. Rdzawski, G. Muzia, S. Pawlak,
Quantitative analysis of the fiber content distribution in CFRP
composites using thermal non-destructive testing, Archives of
Materials Science and Engineering 41 (2010) 28-36.
[23] B. Diveev, A. Smolskyy, M. Sukhorolskyy, Dynamic
rigidity and loss factor prediction for composite layered
panel, Archives of Materials Science and Engineering 31
(2008) 45-48.

References
Acknowledgements